Claude Code proficiency check
How well
do you use Claude?
We scan your installed skills, agents, hooks, and automations and grade how well you use Claude Code. We surface the features you're missing and hand you the exact commands to fix them.
EXAMPLE REPORT
This is a fictional user. Get your own score below in under a minute.
CLAUDE CODE PROFICIENCY REPORT
Total score 58 out of 100, grade B
Solid fundamentals. You're leaking points on automation and delegation
Persona: 6-month user · sample data
Area-by-area check
- Coding basicsOn it85
- CustomizationOn it70
- Skill ecosystemMissing out40
- Agent delegationMissing out30
- Model strategyOn it60
- Browser verificationMissing out20
- Memory & contextOn it55
- Automation & schedulingMissing out10
- OrchestrationMissing out25
- External integrationsNot needed15
What this user should fix now
For each feature they're missing, here's a ready-to-use pick and its install command.
Automation & schedulingMissing out
- /scheduleBuilt into Claude Code — no install
Daily/weekly auto-run routines (cron)
/schedule - /loopBuilt into Claude Code — no install
Conditional repeat and polling automation
/loop 10m <할일>
Skill ecosystemMissing out
- insane-searchInstall
A reader that breaks through blocked sites (Reddit, X, YouTube, etc.)
/plugin marketplace add https://github.com/fivetaku/gptaku_plugins.git 후 /plugin install insane-search@gptaku-plugins - cavemanInstall
Cuts output tokens by 65% — code and errors stay intact
github.com/juliusbrussee/caveman 의 SKILL.md를 ~/.claude/skills/caveman/ 에 복사
Browser verificationMissing out
- Playwright MCPInstall
Claude opens a real browser to click and verify with screenshots
claude mcp add playwright -- npx @playwright/mcp@latest
Now check your score
Get your own report card just like the example above. Just run the scanner locally — that's it.
Run the scanner locally
Read-only · sent only after you consent$node scanner/checkup.mjsClone the repo and run the command above. It shows you exactly what will be collected in the terminal, asks for consent (y/N), then sends the results. When it's done, your results page opens.
Collected: counts and settings flags only. File contents and names are never sent. Your skill list and code never reach the server.
Daily Claude news, free
New skills, updates, and usage tips, curated and sent once a day. You can subscribe without taking the check.
Need a tailored improvement plan and weekly re-checks?
See pricing →